From the article :
"Insiders say that marketing missteps and duplicated development processes meant IBM Cloud was doomed from the start, and eight years after it attempted to launch its own public cloud the future of its effort is in dire straits"
I would say that this is a symptom of a deeper malaise. From following IBM over the last decade or so, I've come to believe they are facing the consequences of the "EPS 20" [0]. They had the EPS 20 for so long that the core R&D and other enablers towards the EPS 20 were sacrificed. Former employees, your views?
Submitter here. Yes, absolutely the case. Throughout the late 2000s, the singular focus was on EPS 20. During that time they still had a deep bench of engineer talent, easily capable of building a capable public cloud. Unfortunately, layoffs were frequent and indiscriminate to performance. It was not uncommon to see entire job categories or departments eliminated.
Then, by the early 2010s IBM began to cry for 'cloud skills'. I never understood this. AWS was not built with 'cloud skills'. IBM had the breadth and depth in software and hardware, squandered it and ultimately failed to produce a cloud the world wants.
Agreed. Huge mistake by Palmisano to focus on the finances (the end result) rather than the products and tech that generated them.
Hindsight is 2020, but I also see it as a huge error that they sold their fabs off. When you look at the innovation and interest in semis these days across mobile and in the cloud, could have ended up as a significant differentiator / value generator for them. Instead they saw it as “empty calories” as Rometty put it; all revenue and no profit. Had they invested in it, could have been a really intriguing offering.
Worked for IBM Cloud right after they started blue-washing Softlayer.
Besides what the author described in the article, there were two IBM Clouds: one for infrastructure (aka Softlayer) and another for PaaS/SaaS (aka IBM Software group's Bluemix) with totally different technologies (although Bluemix Public ran on top of Softlayer) and, as a must from IBM, competing sales teams.
Customers _really_ liked this when they discovered they bought some Bluemix credits but what they really wanted to use were IaaS. It took a long time for the first integration to arrive, and customers got very happy: paying Softlayer in Bluemix credits and paying Bluemix PaaS through Softlayer billing. Yes, billing was (still is?) split, one went through IBM's order systems, the other through Softlayer's.
Of course, there were also a couple of internal competitors for private cloud besides Bluemix Private that competed for customers as well, but let's keep that aside.
Besides competition between IBM Cloud aka Softlayer and IBM Software (who later merged), IBM GTS also had their hosting business, with their own datacenters and managed services (that they sold as "cloud", naturally) that offered IBM Power and Z - both never offered through Softlayer, so customers who wanted everything managed for them (so almost every IBM customer) kept their servers at IBM GTS.
And then there's the kick: after Ginni's mandate, eventually IBM publicly offered Power servers on Softlayer's platform, and there was much rejoicing.
There was just one tiny, little problem: It wasn't IBM Power. It didn't run AIX, just Linux, as Watson ran on Linux. It was closer to a TALOS workstation than to what an IBM customer needed or wanted at all.
Imagine saying to a customer that yeah, we've got Power servers! But no, you can't run AIX, but yeah we are IBM and yeah we still make Power servers. Call our friends at IBM Systems, or IBM GTS if you want that hosted for you.
Those servers weren't IBM at all, except for the CPU. Not even the most locked-in IBM customer would be able to use IBM Cloud if they wanted - and they wanted to.
What, hear the people in the field or the customer? Nah. Crazy times.
Current IBM Cloud employee here . There is a good amount of insider information in this article. Most of it is true. But one big omission is that genesis was actually designed to use a torus network topology with no top of rack switches. Yeah no idea who cooked that idea up. The current issues with the v2 cloud are ibm trying to use kuberneties and other tech because google does . We could do better if we just did our own thing . Also ibm has too many saas tools from outside vendors . For a company who is/was known for software development it’s a shame we have ended up here . One last thing the power cpu offering is being converted from supermicro/tyan to system z ; and will be called ZaaS
9 comments
[ 3.1 ms ] story [ 31.1 ms ] threadI would say that this is a symptom of a deeper malaise. From following IBM over the last decade or so, I've come to believe they are facing the consequences of the "EPS 20" [0]. They had the EPS 20 for so long that the core R&D and other enablers towards the EPS 20 were sacrificed. Former employees, your views?
[0]https://www.barrons.com/articles/ibm-slips-reiterating-20-ep...
Then, by the early 2010s IBM began to cry for 'cloud skills'. I never understood this. AWS was not built with 'cloud skills'. IBM had the breadth and depth in software and hardware, squandered it and ultimately failed to produce a cloud the world wants.
Hindsight is 2020, but I also see it as a huge error that they sold their fabs off. When you look at the innovation and interest in semis these days across mobile and in the cloud, could have ended up as a significant differentiator / value generator for them. Instead they saw it as “empty calories” as Rometty put it; all revenue and no profit. Had they invested in it, could have been a really intriguing offering.
Besides what the author described in the article, there were two IBM Clouds: one for infrastructure (aka Softlayer) and another for PaaS/SaaS (aka IBM Software group's Bluemix) with totally different technologies (although Bluemix Public ran on top of Softlayer) and, as a must from IBM, competing sales teams.
Customers _really_ liked this when they discovered they bought some Bluemix credits but what they really wanted to use were IaaS. It took a long time for the first integration to arrive, and customers got very happy: paying Softlayer in Bluemix credits and paying Bluemix PaaS through Softlayer billing. Yes, billing was (still is?) split, one went through IBM's order systems, the other through Softlayer's.
Of course, there were also a couple of internal competitors for private cloud besides Bluemix Private that competed for customers as well, but let's keep that aside.
Besides competition between IBM Cloud aka Softlayer and IBM Software (who later merged), IBM GTS also had their hosting business, with their own datacenters and managed services (that they sold as "cloud", naturally) that offered IBM Power and Z - both never offered through Softlayer, so customers who wanted everything managed for them (so almost every IBM customer) kept their servers at IBM GTS.
And then there's the kick: after Ginni's mandate, eventually IBM publicly offered Power servers on Softlayer's platform, and there was much rejoicing.
There was just one tiny, little problem: It wasn't IBM Power. It didn't run AIX, just Linux, as Watson ran on Linux. It was closer to a TALOS workstation than to what an IBM customer needed or wanted at all.
Imagine saying to a customer that yeah, we've got Power servers! But no, you can't run AIX, but yeah we are IBM and yeah we still make Power servers. Call our friends at IBM Systems, or IBM GTS if you want that hosted for you.
Those servers weren't IBM at all, except for the CPU. Not even the most locked-in IBM customer would be able to use IBM Cloud if they wanted - and they wanted to.
What, hear the people in the field or the customer? Nah. Crazy times.
https://cloud.ibm.com/catalog/services/power-systems-virtual...
Outsourced GTS stuff is going to Kyndrl
Finally! Happy to hear it launched!
> Outsourced GTS stuff is going to Kyndrl
Still, IMHO the spin-off was too late. Now... let's do GBS :)